Pricing: what to charge and how to charge for it

You can lose half a day changing $39 to $49, checking three competitors, changing it back, then avoiding the pricingThe decisions covering how much a customer pays, what creates the charge, what is included, and when payment is due. page because the number still feels made up. And at this stage, part of it is; putting the number in a tasteful font feels better, but it won't make it more real.

Checking more competitor pages won't tell you what your price should be. You need a number you can explain; it should seem reasonable for the result your customer wants, cover what it costs you to deliver, and be tested with somebody who can choose to pay or walk away.

Your first price is a proposal

I think you should set and test a price from zero customers; waiting until you know what people will pay stops you from finding out.

This doesn't mean choosing a random number. Your first price should have a reason behind it, and you should know which parts are still assumptions. “Would you pay $50 for this?” lets somebody talk about make-believe money. A payment, deposit, signed paid pilotA limited early use of a product or service, usually with agreed scope, length, participants, and measures. A pilot may be paid or free; each provides different evidence., or refusal of a real offer tells you what they did when they could pay or walk away.

A price has three jobs

One number has to make sense to the customer, cover your costs, and fit the way they buy. Those are three separate questions:

Start with their comparison

Your customer won't look at the number alone. They'll compare it with what they already pay, the time and money the problem costs them, a manual workaround, another product, hiring somebody, or leaving the problem alone. Whatever they use for that comparison is called their reference priceThe amount or alternative a customer uses to judge whether your price seems high, low, or reasonable. It may come from another product, a workaround, doing nothing, or what they expected the category to cost..

Earlier chapters should give you some idea of the result they want, what the problem has already cost them, what they do now, and what they're afraid of getting wrong. If customers have shown that the problem costs them $500, you can explain why you're testing $50; if you invented the $500, you still have an assumption to test.

Competitor prices can help if you compare like with like. Record what each competitor charges, who the price is for, what the customer receives, and when you checked it; then compare that with your offer.

If you're using the result a customer expects to help set the price, that's value-based pricingSetting and testing a price with reference to the result the customer values. Cost and competitor prices remain separate inputs.. You still need to check that the price covers your costs. (Stripe, August 2025)

What creates the charge

You also need to decide exactly what you're charging customers for. It might be one project, one month of access, each person who uses the product, each booking, or an agreed result. This is your pricing modelThe rule that decides what the customer pays for, such as access, each user, each booking, usage, a completed result, or a transaction., and it needs to make sense to the customer, cover the costs that rise as they use more, and let them work out what they'll pay.

Software with high AI or API costs needs particular care. One heavy user can cost more than a flat monthly fee, while per-use pricing can leave the customer unable to predict the invoice; a monthly allowance, or a base fee plus usage, can stop one customer wiping out your margin without giving them an open-ended bill. (Stripe, April 2026). Make sure you set concrete per-user AI API guardrails before you launch.

The money left after the sale

A sale can bring in revenue and still leave you worse off if it costs too much to deliver. Materials, hosting, AI use, payment processing feesThe amount charged by a payment provider to process a transaction, often made up of a fixed amount, a percentage, or both., support, delivery labor, refunds, and shipping are costs tied directly to a sale. These are your cost of goods sold (COGS)The direct cost of producing or delivering the goods or services sold during a period. It's also called cost of sales; accounting treatment varies by business and jurisdiction. or cost of sales.

Subtract those direct costsA cost caused by producing, delivering, or supporting a particular sale, customer, unit, or transaction. from revenue, and you get gross profitRevenue left after subtracting cost of goods sold or cost of sales.; divide gross profit by revenue, and you get gross marginGross profit expressed as a percentage of revenue.. After marketing, admin, product development, software, and salaries, what remains is operating profitGross profit left after operating expenses have been deducted.; as a percentage of revenue, that's your operating marginOperating profit expressed as a percentage of revenue.. Net profitRevenue remaining after direct costs, operating expenses, and other expenses such as interest and tax. includes the remaining items, such as interest and tax. (U.S. Securities and Exchange Commission, accessed September 2026)

If you charge $50 and the direct cost is $10, gross profit is $40 and gross margin is 80%. That still leaves the costs of finding the customer, building the product, keeping the accounts, and paying yourself. A solo service can appear wildly profitable when you've priced your delivery time at $0 (an excellent rate for everybody except you).

There isn't one good margin for every kind of business. The figures below show how much margins vary across established companies; a solo business may have very different numbers.

(TLDR; don't start a grocery store)

These figures describe U.S. public companies in January 2026; the operating column uses Damodaran's pre-tax unadjusted measure, and solo businesses have different costs. SaaS Capital reports 80% to 85% gross margin on license revenue across its portfolio and survey data. (Damodaran, January 2026; SaaS Capital, 2025)

To work out whether one customer, booking, or sale pays for itself, put its revenue beside all the costs it creates, including your time. These are your unit economicsThe revenue and costs attached to one customer, sale, product, booking, transaction, or another unit that makes sense for the business..

How they reach payment

Once somebody is interested, what has to happen before they can pay? They may check out, start a trial, watch a demo, agree to a paid pilot, accept a proposal, or go through procurementThe process an organization uses to check, approve, contract, and pay for a purchase.. That path is your sales mechanismThe route a customer follows from interest to payment, such as self-serve checkout, a trial, demo, paid pilot, proposal, or procurement process..

If the offer is easy to understand, set up, and reverse, the customer may be able to buy through self-serve checkoutA purchase route in which the customer can understand the offer, choose it, pay, and begin without speaking to the seller.. A live product demoA recorded, interactive, or live presentation showing how a product works. A live demo can also help a buyer check fit and ask questions before purchase. makes sense when they need help with fit, implementation, security, or approval; a paid pilot lets them test a larger decision in their own environment.

Gartner's August and September 2025 survey of 645 B2B buyers found that 67% preferred a sales-rep-free experience and 69% wanted to check AI-generated information with a rep. The same buyer can want to research alone and speak to somebody before committing. (Gartner, May 2026)

Requiring a demo makes sense when buying without one could leave the customer with the wrong setup, a difficult implementation, or no way to answer an approver's questions.

Three prices

If different customers need different amounts of access, volume, speed, support, rights, or risk cover, separate plans can help them choose. This is when you'll introduce pricing tiersOne plan within a set of offers, with a defined price, customer, result, scope, usage, access, or support level., and three is a common number of choices you'll see - there's a reason for this (lots of it is based in psychology).

First, when a smaller and larger choice are presented to a user, a middle option can feel easier to justify: researchers call this the compromise effectThe increased appeal a middle option can receive when it sits between a lower and higher option on the qualities being compared.. In one preregistered 2026 television study, 58.4% chose the middle option in a three-tier set, compared with 47.6% when the smaller option was absent; the effect changed when each level contained several choices. It wasn't a SaaS pricing page. (Journal of Consumer Research, 2026)

Second, too many choices can make a customer less likely to choose at all. In Iyengar and Lepper's 2000 grocery-store experiment, 60% of passersby stopped at a display of 24 jams, compared with 40% at a display of six; among those who stopped, nearly 30% of the six-jam group bought a jar, compared with 3% of the 24-jam group. The idea that a large set of similar options can make a decision harder is usually called choice overloadA possible effect in which a large number of options makes choosing harder, less satisfying, or less likely. Research results vary by setting; the number of choices alone doesn't predict what customers will do.. (Iyengar and Lepper, 2000)

The wider evidence is mixed. A 2010 meta-analysis of 63 conditions from 50 published and unpublished experiments, involving 5,036 people, found an average effect close to zero and considerable differences between studies. The researchers couldn't identify a reliable rule for when more choice becomes too much. I wouldn't use the jam study to argue that three prices always beat five; it's a reason to check whether each plan helps your customer choose or gives them another comparison to make. (Scheibehenne, Greifeneder, and Todd, 2010)

The number of plans should follow the buying routes and situations you have available. A narrow customer with one result may need one plan; if there are only two real routes, two are enough. A plan included only to make another look better is a decoy, and researchers call the change it causes the decoy effectA change in preference caused by adding an option that is clearly worse than another available option on the relevant qualities.. Each plan should be one a real customer could realistically choose, and a “most popular” badge should ideally be based on real sales.

What the number says

The last number of your price means more than you think:

The first finding comes from five experiments by Thomas and Morwitz, the second from five studies by Wadhwa and Zhang; the precise-number research tested opening offers in negotiations and didn't test website prices. (Thomas and Morwitz, 2005; Wadhwa and Zhang, 2015; Mason et al., 2013)

When $49.99 feels lower than $50 because the first digit changes, that's the left-digit effectThe tendency for a just-below price to seem lower when it changes the left-most digit, such as $49.99 compared with $50., and prices that end below a round number are often called charm pricingPricing that ends below a round number, commonly in 9, such as $49.99.. A first number influencing the negotiation is price anchoringThe effect of an initial number or price on later judgments and comparisons..

There isn't a fixed line between good value and cheap. A low price can reduce risk for a familiar product, and it can also make a buyer question how you'll provide an expensive result for so little. Where buyers can't judge quality before purchase, some use price as a clue. This is called a price-quality cueThe information a customer may infer about quality, seriousness, or intended buyer from the price when quality is difficult to judge before purchase.. (Rao and Bergen, 1992)

The number has to be believable alongside what you promise and the work required to deliver it. Changing $50 to $49.99 won't explain how a $10 service provides a result that appears to require three hours of expert work.

Free changes the evidence

A free user shows willingness to try; a free trialTime-limited access that lets a potential customer use a product or service before deciding whether to pay. tests whether they can reach a result, while freemiumA model with continuing free access to a limited version and paid access to additional features, capacity, support, or use. keeps limited access free. A discounted customer proves payment at the lower price; a paid pilot also tests delivery.

Across 200 B2B software products in ChartMogul and ProductLed's 2026 research, median free-to-paid conversion was 8%; card-required trials converted more than five times as often as no-card trials and asked for a stronger commitment at entry. Requiring a card produces a different group of trial users. (ChartMogul, March 2026)

An early discountA reduction from a stated full price, usually tied to a reason, customer, volume, commitment, or period. needs a reason, an end, and a recorded full price. “Founding customer rate for three months in return for weekly feedback” explains why you offered the lower price; 40% off forever can make the first experiment rather long.

Example: pricing a booking platform

Imagine you're selling a booking platform to independent dog groomers. It takes a customer's deposit when they book an appointment, and you're deciding whether the groomer should pay $39 a month to use it or $1 for every booking confirmed through the platform. For this example, you expect a typical groomer to receive 50 bookings a month; each account costs $7 a month to host and support, plus $0.12 for every booking processed.

At 20 bookings, the subscriptionA recurring payment for continuing access to a product or service. would leave you with $29.60 gross profit and a 75.9% gross margin; charging per booking would leave you with $10.60 and a 53% gross margin.

The subscription gives the groomer a predictable billA price a customer can estimate, understand, and limit before receiving the invoice. and gives a heavier user better value. Per-booking pricing lowers the commitment for a lighter user and lets your revenue rise with activity. A hybrid $19 + $0.50 per booking sits between them, although adding it creates a slightly tougher need for pricing articulation and another billing rule.

A spreadsheet can't show which option groomers prefer. A small cohort accepting or refusing what you're offering is much more helpful.

“Too expensive” is incomplete feedback

“Too expensive” tells you they won't pay under the current terms; it doesn't tell you why. They may mean the number is too high, they don't have the budget, the result or proof is unclear, the timing is wrong, the bill is unpredictable, or somebody else has blocked the purchase. You can ask what they compared it with and what would have to be true for the price to make sense. Some answers will still be no.

The pricing record gives you somewhere to keep one price, model, calculation, buying route, and paid test. It also records acceptance, refusal, discount requests, use, support, renewal, churnWhen a customer stops paying for or using a recurring product during a chosen period., and referral. You can keep compliments and future intent in the same record without counting them as payment.

Review when delivery cost, your time, usage, customer group, product scope, renewal, churn, or capacity changes. A price increase needs a reason, clear notice, and a decision about existing customers; changing the amount, plans, and audience together makes the result impossible to read. (Stripe, accessed September 2026)

You're finished for now when a customer can understand and pay the proposed price, your calculation includes the costs you know about, and a review date is in the record. The next chapter, Distribution fundamentals, uses that price to work out which ways of reaching customers the business can afford.

Prompt: build my first pricing test

The prompt below can go into the model you use, along with your customer notes, current alternatives, competitor pages, costs, and any payment or refusal evidence; “I don't know yet” is fine where the information doesn't exist.

I need to choose a first price and charging model, check whether I can afford to
deliver it, and create one test in which a real customer can pay or refuse.

Purpose
Turn my customer and cost information into a pricing proposal I can test. Keep
facts, estimates, assumptions, missing information, and recommendations
separate.

When I'm using it
[idea / prototype / lightweight MVP / pre-launch / live]

My context
- Product or service: [what I'm selling now]
- Customer group: [who I'm concentrating on]
- User, buyer, and approver: [same person / different people / I don't know]
- Result I help them get: [result supported by my research]
- Current alternatives: [products, people, manual work, AI, or doing nothing]
- Cost or consequence of the problem: [customer evidence, or I don't know]
- Positioning: [my current one-sentence explanation]
- Proof I have: [results, demo, pilot, references, or other proof]
- Competitor evidence: [price, package, customer, URL, and date checked]
- Prices or models I'm considering: [up to three, or I don't know]
- How customers may buy: [checkout, trial, demo, pilot, proposal, procurement]
- Direct cost per customer or sale: [materials, hosting, AI/API use, fees,
  support, delivery labor, contractors, refunds, shipping, or other]
- My delivery time: [hours and the hourly or replacement-cost assumption]
- Monthly operating costs: [software, admin, product, marketing, salaries, other]
- Capacity: [customers, bookings, projects, or hours I can support]
- Payment and behavior evidence: [paid, refused, requested discount, used,
  renewed, canceled, referred, or none yet]
- Limits for this test: [time, money, customer count, or delivery capacity]
- Next review date: [date]

Your task
1. Extract only the facts and numbers I've supplied. Keep each source and date.
2. List the assumptions, missing figures, and unsupported customer results.
3. Explain what the customer may compare the price with, using only my
   evidence. Include the current arrangement and doing nothing.
4. Compare no more than three charging models. For each, show what creates a
   charge, how the customer receives value, how my costs may change, how
   predictable the bill is, and the main problem to watch.
5. Recommend one model for the first test. Explain the decision and show which
   part remains uncertain.
6. Calculate gross profit and gross margin from my figures. Estimate operating
   profit and operating margin only when I supplied enough information. Show
   every formula and don't treat my time as free.
7. Recommend one, two, or three plans. Use three only when my evidence supports
   three different buying situations; don't invent a decoy or label a plan
   “most popular” without real data.
8. Recommend a number presentation, such as round, just-below, or calculated
   proposal. State the research limit and don't claim that an ending will raise
   sales.
9. Recommend a route to payment: self-serve, trial, recorded demo, live demo,
   paid pilot, proposal, or procurement. Base it on the customer's risk and the
   work needed to succeed. Don't copy another company's process without that
   evidence.
10. Create one test in which a matching customer can accept or refuse the real
    price. Include the offer, customer group, route, number of attempts, evidence
    to record, stop rule, and review date.
11. Give me five follow-up questions for a customer who says “too expensive.”
    The questions should explore their comparison, budget, value, proof, timing,
    and approval route without arguing with them.
12. Set a review rule using payment, refusal, discount requests, usage, support,
    renewal, churn, capacity, or cost changes.

Rules
- Don't invent willingness to pay, customers, quotes, results, competitor
  prices, conversion, usage, costs, margins, benchmarks, or tax treatment.
- Don't turn a competitor average or industry margin into my recommended price.
- Don't claim that a higher price signals quality in every market.
- Don't claim that three plans, charm pricing, anchoring, a free trial, or a demo
  will increase conversion.
- Don't treat a free user, discounted customer, pilot, and full-price customer
  as the same evidence.
- Don't recommend outcome pricing unless the result and attribution are clear.
- Don't recommend usage pricing without showing how the customer can predict or
  limit the bill.
- Don't give legal, tax, or accounting advice.
- Use American English and ordinary language; explain a commercial or finance
  term before using it.

Return exactly these headings
## Facts and sources
## Assumptions
## Missing information
## What the customer compares the price with
## Charging-model comparison
## Recommended first model
## Cost and margin calculation
## Number of plans
## Number presentation
## Route to payment
## Real-payment test
## Questions after “too expensive”
## Review rule

You'll still need to check the result against your source material. A model can calculate the margin from the numbers you give it; it can't discover a customer's willingness to payThe highest amount a customer is prepared to pay for a particular offer under particular conditions. A stated amount remains weaker evidence than a real payment decision. from a polished description of the product.

Sources

• U.S. Securities and Exchange Commission. “Beginners' Guide to Financial Statements.” Accessed 14 September 2026.
• Damodaran, Aswath. “Margins by Sector (US).” Data as of January 2026. Accessed 14 September 2026.
• SaaS Capital. “What Should Be Included in COGS for My SaaS Business in 2025?” Accessed 14 September 2026.
• Thomas, Manoj, and Vicki Morwitz. “Penny Wise and Pound Foolish: The Left-Digit Effect in Price Cognition.” Journal of Consumer Research, June 2005.
• Wadhwa, Monica, and Kuangjie Zhang. “This Number Just Feels Right.” Journal of Consumer Research, February 2015.
• Simonson, Itamar. “Choice Based on Reasons: The Case of Attraction and Compromise Effects.” Journal of Consumer Research, September 1989.
• Sharif, Marissa A., Elizabeth C. Webb, and Sudeep Bhatia. “Extremeness Aversion and Choice Set Composition.” Journal of Consumer Research, August 2026.
• Iyengar, Sheena S., and Mark R. Lepper. “When Choice Is Demotivating: Can One Desire Too Much of a Good Thing?” Journal of Personality and Social Psychology, December 2000.
• Scheibehenne, Benjamin, Rainer Greifeneder, and Peter M. Todd. “Can There Ever Be Too Many Options? A Meta-Analytic Review of Choice Overload.” Journal of Consumer Research, October 2010.
• Mason, Malia F., Alice J. Lee, Elizabeth A. Wiley, and Daniel R. Ames. “Precise Offers Are Potent Anchors.” Journal of Experimental Social Psychology, July 2013.
• Rao, Akshay R., and Mark E. Bergen. “Price Premium Variations as a Consequence of Buyers' Lack of Information.” Journal of Consumer Research, December 1992.
• Stripe. “SaaS Pricing and Packaging Strategy.” Updated 7 April 2026.
• Stripe. “AI Pricing Models.” Updated 19 April 2026.
• Stripe. “Value-Driven Pricing.” Updated 11 August 2025.
• Stripe. “Pricing Strategies for New Products.” Accessed 14 September 2026.
• ChartMogul and ProductLed. “The Conversion Report.” 18 March 2026.
• Gartner. “69% of B2B Buyers Turn to Sales Reps to Validate AI-Generated Insights.” 20 May 2026.